- Joined
- Sep 26, 2021
- Posts
- 6,115
- Main Camera
- Sony
I got my first Mac when Exponential Technology went out of business, and they gave away test machines to some of the last remaining employees. (I stayed on to the very end, two months after they stopped paying me. I did eventually get paid though).
I joined Exponential Technology in 1996, just after their first tapeout. I was quickly handed half the chip (by area, at least), because my PhD research had involved exactly the same weird types of circuits that Exponential was using. To this day I recall sitting in my first interview of the day there with Cheryl, who sat cross-legged on her chair which was turned backwards, and she started to draw a couple of circuits and a dashed line between them. I asked her “are you going to ask me about level-shifting?” and she said “yeah, you’re hired. Let’s go meet the rest of the team.” She and I would work together for many more years at AMD, but that’s not a Mac story.
We were aiming for 533 MHz, which, at the time, would have been the highest clock on earth (back when people cared about clocks), but shortly after I got there they got first silicon back and found they were only hitting a clock rate somewhere in the 400’s. I remember we printed out schematics on huge sheets of paper and started trying to figure out what was causing the unexpected critical path, doing the Roth D algorithm to figure out how to propagate the necessary signal transitions to where they needed to go. Turned out the problem was that, in order to reduce power consumption, they wrote a tool called ”Hoover” (before i got there) and used it to automatically reduce drive strengths on various gates. This caused a cross-coupling problem, which I wrote about in the JSSC paper referenced below. This was probably the seed that got me interested in electronic design automation, which became an accidental theme of the rest of my career.
Anyway, we fixed the problem and were getting ready to ship production chips to Apple clone-makers, but we had a problem. To boot the Mac’s OS using our chip, we had to make changes to the BIOS. Steve Jobs came back, and permission to change the BIOS went away (even though Apple was an investor).
Internally, we had working Macs that had been hacked to use our chips. We also had Windows NT running on our chip, and it smoked Mac OS at the time (in terms of speed). There were test rigs (Macs and such) in the mysterious back room of the offices down on Zanker Road In San Jose. One night, Allan and I were working late, and at midnight a fire alarm went off back there. I had to call a manager or two (because nobody was in the office and I didn’t even have the key for back there), so that was fun.
I remember reading this story about the company’s demise in 1999 and discussing with some coworkers, most of whom agreed it was a little too favorable to management. Still a good background on the thing.
Here’s the paper we wrote for IEEE’s Journal of Solid State Circuits (the journal, btw, that my PhD advisor was desperate for me to get published in when I was a grad student). I took the lead on it but a lot of it was written by other people, and a lot of what I did was merely editing. The photo of me in the back makes me giggle.
Some of the folks from our office moved to Austin and formed a second design team. The Exponential PowerPC was originally supposed to have dual personalities, running both x86 and PowerPC instructions. My recollection is that the texas team was going to rekindle that (or switch to x86?) without getting sucked into the day-to-day need to support the ”shipping” product up in California. When we went under, that texas team became EVSX (“everything else sux”), which then became Intrinsity, which then became a good chunk of Apple’s CPU design team.
I powered up my Mac a couple of times, realized I didn’t really have any use for it, and stuck it in the garage. Right around the time Leopard was released, I let go of my grudge against Steve Jobs for taking away the best job I ever had, and picked up a 17” MacBook Pro. I was interested in Tiger, but when I was in law school we had to run software called Examsoft, and only PC’s were an option back then.
Since then it’s been nothing but Mac after Mac - I have a stack of ‘em sitting in a closet. It’s hard to believe that something that started that long ago is still a thing.
I joined Exponential Technology in 1996, just after their first tapeout. I was quickly handed half the chip (by area, at least), because my PhD research had involved exactly the same weird types of circuits that Exponential was using. To this day I recall sitting in my first interview of the day there with Cheryl, who sat cross-legged on her chair which was turned backwards, and she started to draw a couple of circuits and a dashed line between them. I asked her “are you going to ask me about level-shifting?” and she said “yeah, you’re hired. Let’s go meet the rest of the team.” She and I would work together for many more years at AMD, but that’s not a Mac story.
We were aiming for 533 MHz, which, at the time, would have been the highest clock on earth (back when people cared about clocks), but shortly after I got there they got first silicon back and found they were only hitting a clock rate somewhere in the 400’s. I remember we printed out schematics on huge sheets of paper and started trying to figure out what was causing the unexpected critical path, doing the Roth D algorithm to figure out how to propagate the necessary signal transitions to where they needed to go. Turned out the problem was that, in order to reduce power consumption, they wrote a tool called ”Hoover” (before i got there) and used it to automatically reduce drive strengths on various gates. This caused a cross-coupling problem, which I wrote about in the JSSC paper referenced below. This was probably the seed that got me interested in electronic design automation, which became an accidental theme of the rest of my career.
Anyway, we fixed the problem and were getting ready to ship production chips to Apple clone-makers, but we had a problem. To boot the Mac’s OS using our chip, we had to make changes to the BIOS. Steve Jobs came back, and permission to change the BIOS went away (even though Apple was an investor).
Internally, we had working Macs that had been hacked to use our chips. We also had Windows NT running on our chip, and it smoked Mac OS at the time (in terms of speed). There were test rigs (Macs and such) in the mysterious back room of the offices down on Zanker Road In San Jose. One night, Allan and I were working late, and at midnight a fire alarm went off back there. I had to call a manager or two (because nobody was in the office and I didn’t even have the key for back there), so that was fun.
I remember reading this story about the company’s demise in 1999 and discussing with some coworkers, most of whom agreed it was a little too favorable to management. Still a good background on the thing.
Here’s the paper we wrote for IEEE’s Journal of Solid State Circuits (the journal, btw, that my PhD advisor was desperate for me to get published in when I was a grad student). I took the lead on it but a lot of it was written by other people, and a lot of what I did was merely editing. The photo of me in the back makes me giggle.
Some of the folks from our office moved to Austin and formed a second design team. The Exponential PowerPC was originally supposed to have dual personalities, running both x86 and PowerPC instructions. My recollection is that the texas team was going to rekindle that (or switch to x86?) without getting sucked into the day-to-day need to support the ”shipping” product up in California. When we went under, that texas team became EVSX (“everything else sux”), which then became Intrinsity, which then became a good chunk of Apple’s CPU design team.
I powered up my Mac a couple of times, realized I didn’t really have any use for it, and stuck it in the garage. Right around the time Leopard was released, I let go of my grudge against Steve Jobs for taking away the best job I ever had, and picked up a 17” MacBook Pro. I was interested in Tiger, but when I was in law school we had to run software called Examsoft, and only PC’s were an option back then.
Since then it’s been nothing but Mac after Mac - I have a stack of ‘em sitting in a closet. It’s hard to believe that something that started that long ago is still a thing.
